An Adaptive Framework for Designing Secure e-Exam Systems

摘要

The most remarkable feature of the Internet during the last few years has been the fast propagation of social networking platforms. These platforms allow users to communicate with each other and share information. Consequently, tens of thousands of messages are generated every second on social networks. Nevertheless, several security threats exist in these networks of which spam messages are considered the most prominent. Therefore, a great deal of research has been conducted to detect such messages. However, Arabic research is still limited. Thus, in this research, we proposed a new Arabic spam detection system that combines the Rule-Based scoring technique with the Na?ve Bayesian classifier to detect spam messages in Arabic that is specifically targeting Saudi Arabia users of social networks. After gathering and analyzing the dataset, we chose three content-based features that can distinguish spam messages from legitimate messages. Based on our experimental results, we showed that the Rule-Based scoring technique achieved 52% accurate detection results, while the Na?ve Bayesian classifier achieved 86% accurate detection results.